Friday, 31 July 2015

Forecasting, when Modelling is not the only Choice - Forecasting and Virtualization (5 of 5)

Virtualization has thrown us a lot of curve balls when it comes to forecasting. It’s a compact environment with a smaller footprint which reduces your IT costs.

It always throws up the questions how much more can I fit? How much headroom do I have?

What is the Correct Technique?
This is the big question and which technique you choose is dependent on:
        What the end-user is expecting.
        Which technique gives you the best result for the question/s that have been asked.
        How much time you have to produce a result.
To summarize my series in a few words you need to:
        Understand what the forecast is to accomplish.
        Know what enterprise standards dictate.
        Ensure you have as much business and technical information as possible.
        Trend when trends make sense.
        Have a repeatable process.
        Document process and assumptions.

I hope you've enjoyed my series and if you have any questions I'll be happy to answer them.
I'll be running a webinar 'Why do we model in the UK, monitor in Japan, and manage in the USA?' - don't forget to register for your place now.

Charles Johnson
Principal Consultant

Wednesday, 29 July 2015

Forecasting, when modeling is not the only choice - Trend Types(4 of 5)

There are three different trend types and the closer the gap between the confidence lines the more confidence you can have in the result.  Setting thresholds also assists in gaining visibility into the trend impact. 

Normal Trend

Step Trend
This takes in to account a change in the business, which you are aware will be happening.

Including the step will allow you to see how much impact that change is going to have on the overall trend.
In this scenario it has little or no effect.
Average vs. Peak Trend
I am often asked the question do I need to look at the averages or do I look at the peaks? You can produce a chart like the one below which looks at both.
The red line in this example is the peak trend and the green line the average trend, this tells me that the peak trend moving forward is  not as significant as the average trend.
So when I start analyzing this environment I’ll choose to look at the averages rather than the peaks.
On Friday I'll take a look at Forecasting and Virtualization and summarize the series for you. In the meantime join our Community and get access to our on-demand webinars, papers and performance tips
Charles Johnson
Principal Consultant


Monday, 27 July 2015

Forecasting, when modeling is not the only choice - Forecasting Techniques (3 of 5)

In my opinion there are four main techniques that are used in forecasting: analytical modeling, simulation modeling, trending and headroom charts.

Let’s take a look at each of these:
Analytical Modeling                                        

        The ability to predict environment changes based upon the arrival of work
·      CPU
·      IO
        Based upon
·      Baseline timeframe (seasonal peaks and troughs, highest transaction times etc)
·      Calibration of model – Does it match real life?
·      What-If changes to model
        Spend virtual $’s, helping you to determine what is the best cost benefit that you can get out of it
        Low cost modeling technique, as you’re able to take the data and apply it going forward without making costly mistakes.
The graph below illustrates how this can be considered, always taking in to account how the ‘end user’ will be affected.
Simulation Modeling
There are many organizations such as Wall Street companies, who feel the need to run simulation models but it does take longer to set up and can be very costly.
        Replication of environment
        Has to stay close to production environment
        Usage of tools to replicate workload into simulated environment
        Longer lead time to setup
        Provides granular detail on environment changes
        Can be costly
With trending you are usually looking at one metric for each chart, this doesn’t prohibit you from having several charts looking at different metrics for the same scenario.
        Provide prediction based upon an interval of time
        Quick to produce
        Usually looking at one metric
        Can show steps in growth changes
        Confidence is based on length of interval and amount of historical data
The more data that you have to feed in to your trend chart then the more confidence you can have in the end result.
On Wednesday I'll be taking a more in-depth look at Trend types.
Charles Johnson
Principal Consultant

Thursday, 23 July 2015

Forecasting, when Modelling is not the only Choice - Forecasting Challenges(2 of 5)

Today I'll be looking at the challenges we face when forecasting. You need to find the right balance.

The impact on the service versus the impact of the cost is something that must always be taken in to consideration. A prime example of this is Cloud Computing, which has helped some organisations to find this right balance. Utilizing Cloud resources efficiently means knowing how much you are going to need and when, you still need to forecast your requirements in advance if you don’t want costs to spiral out of control.

Forecasting relies on:

· Proper technical and business data

· Proper business information - what is the business doing overall, is there growth, decline, what is new

· Valid technical and business assumptions 

· Ability to compare past activity
· Match needs vs. cost

Forecasting Scenarios
I like to put these in three different categories:
• Near term – 3 to 12 months
• Long term – 1 to 3 years, 1 to 5 years
• Environment changes
                   Downsizing Infrastructure
                      Right sizing Infrastructure
                          IT mergers and acquisitions

Where do we get the data?

Many organizations have a variety of platforms, there’s all types of data available.

• Data collected from various point tools or Capacity Management software.
• Past forecasts, enable you to get closer and closer to accuracy.

• Business users and owners

        May want to help

           May say “You’re the capacity team, you figure it out”

On Monday I'll be taking a look at Forecasting techniques. Why not join our Community and download my recorded webinar on this subject in the meantime

Charles Johnson
Principal Consultant

Tuesday, 21 July 2015

Forecasting – When Modeling is not the only Choice (1 of 5)

What is the definition of Forecasting?

In this series I’m going to take a look at forecasting and how we can set about doing it. So it’s relevant to start with the question what is the definition of forecasting?
Forecasting is “The process of making statements about events in the future”
You need to know what you are going to do in the future, what is going to change in the future.

Why do we forecast?

The underlying question through all of this is:

What are we trying to accomplish? -  We’re trying to look at what is going to happen in the future, particularly where there are going to be changes in the organization. The business may be adding an application, reducing the use of an application or even merging a datacentre. Our job is to make sure that we have the right amount of resources available.

Forecasting Objectives

As IT demand has been growing so has the complexity of our IT systems. In order to make good forecasts we have to:

·  Make business decisions
·  Resolve possible business and IT conflicts
·  Provide IT Efficiency
·  Balance Cost vs. Service

On Thursday I'll be looking at the challenges that forecasting creates for those managing IT capacity and don't forget to come along to our webinar tomorrow 'Data Correlation  for Capacity Management'

Charles Johnson
Principal Consultant

Thursday, 16 July 2015

Capacity Management: Operational or Strategic? (4 of 4)

My previous three blogs in this series have defined two forms of capacity management: operational and strategic.  
Unfortunately two quite different activities, with different techniques and objectives, often get confused and thought of as the same thing.  One name, two activities - it can be confusing. 

Effective capacity management for an organization means one needs to understand both activities and ensure all bases are covered.

Operational capacity management sits much happier as a silo-based activity.  System managers need specialist tools tuned to each environment they manage.  Performance data captured for each platform can tell them what capacity issues are hitting or about to hit the system, so they can address the problem and ensure good performance.   This relies on the same data as strategic capacity management, but often needs to be looked at in more detail.  It’s all about the ‘here and now’, ensuring that today’s users are getting the quality of service they require.

Strategic capacity management addresses different but equally valid needs.  Cost savings tend to be less numerous but of higher value.  As a strategic activity it is better run as a central infrastructure function, cutting across silos.  Following ITIL good practice guidelines, it suits a broader service view, with closer ties to business issues such as new service rollout, services and applications that span across multiple silos and step change such as merger and acquisition.  Its concern is less about whether a given VM will need more resource tomorrow, more about if the current investment in Cloud capacity will meet business needs over the next six months.  It uses the same data as operational capacity management, but with less focus on day to day variations, more on longer term trends.  Its concern is ensuring that as the business evolves, users will get the quality of service they require next week, month or year.

Unsurprisingly, organizations, particularly those implementing newer silos, tend to go for operational capacity management first.  In an earlier blog in this series, I likened the need for capacity management to having a stone in your show.  If you can already feel the stone in your shoe, you need it removed straight away.  As operational capacity management gets addressed first, this often this leads to strategic capacity management being encouraged to use operational capacity management tools to meet their needs.  They are already there and collecting the same data, so why not?

The reason is that this never works, as no point solutions offer the broader view that is necessary to strategic capacity management teams. 

Both approaches can feed off the same data, but different approaches, skills and tools are needed to address their respective questions.  

Not reassessing what your objectives for all capacity management are and what tools will deliver the answers you need means that strategic capacity management often fails to be properly addressed.  This can be painful, as all the stones get into your shoe and have to be removed.  

Fighting each problem like this costs your business time and money.  

Understanding the need and getting the right tools in place for the right job ensures short and long term planning needs are catered for and user satisfaction enhanced over time.
The best approach is to understand what you want to do, and then get the right tools to do the job.

Our ‘Data correlation for capacity management’ webinar takes place on July 22, don't forget to register.

Andrew Smith
Chief Executive Officer

Tuesday, 14 July 2015

Capacity Management: Operational or Strategic? (3 of 4)

In my previous blog in this series I outlined how valuable operational capacity tools are at addressing what are essentially performance concerns related to capacity, for example, how many more VMs can my Host support?  

Such products do not provide a complete picture.  If this is your only capacity management work, it risks missing broader and longer term capacity issues which when addressed  will also have direct financial benefit to your business.

The other day I read an article comparing capacity management to walking with a stone in your shoe.  Operational capacity management is taking the stone out of your shoe either when it has started to hurt you, or when you first feel it is there but before it has caused you any real pain.Much better practice is to fasten the shoe so well that the stone never gets in there in the first place.  

This is strategic capacity management – taking action well in advance to remove the risk that the problem ever occurs.  

No matter how well you tie the shoe, you won’t be able to keep every piece of grit out but you will prevent the larger and nastier stones from getting in there and surprising you.

Not all events we need to plan for can be based on what is happening on the system at the moment.  Just capturing current performance metrics and basing capacity decisions on them is not enough.  We might assess how many more VMs a host can support based on past measurements but in the case of a change to our web site this may change the profile of how users interact with our systems. Capacity could be consumed in ways in which past resource profiles no longer serve as a guide. 

Another example is merger and acquisition, here you need measurement of performance and this can involve bringing together metrics from many disparate systems, across what have up to that point been separate businesses.
What you need then is something that lets you predict capacity based on something other than past performance.  Strategic capacity management needs that past performance data when it is relevant – often much can be learned from past trends in system usage.  

Extrapolating forward can tell us something about how busy our systems will be in the future.  Bringing in business level inputs means that more than trends and extrapolation is required, for example the capability to introduce step change to trends or analytically model how service levels will change over time.

Strategic capacity management tools like Metron’s athene® provide this greater breadth of planning functionality.  This supplements day to day performance orientated operational capacity management activity using point solutions.

One phrase – capacity management, but two differing definitions and deliverables.

On Thursday I'll summarize what I have been saying about operational and strategic capacity management.

Andrew Smith

Chief Executive Officer

Friday, 10 July 2015

Capacity Management: Operational or Strategic? (2 of 4)

In my previous blog I defined two types of capacity management, operational and strategic.

The capacity management promoted by so many point tools for technologies such as VMware is definitely operational.  Performance metrics are collected and analysed, usage of current capacity is assessed and then recommendations are made.  This could be a statement of how many more VMs a host can support, or what workloads should be moved where to avoid running out of capacity. 

Operational capacity management sits much happier as a silo-based activity.  Each silo such as VMware, networks or storage, benefit from tools developed and tuned to their own environment. 

Such tools enable small and highly specific recommendations to be made to tune a system to avoid short term capacity issues.  Such tools are more likely to integrate with that core silo technology, enabling automated change or parameters or movement of workloads.  Cost savings will accrue in small amounts through many small actions.

Undoubtedly these are capacity issues.  Also undoubtedly they are operational issues: short term fixes to problems that are about to occur based on what is known of the environment at that point in time.  

This is valuable work and worthy of implementing specialist solutions.  Nowadays everyone has heard people say ‘Well, think of the cost if the web site is down for a minute’.  Operational capacity management tools are an essential component in making sure such eventualities don’t happen.

Don’t be fooled into thinking that such point solutions are a complete capacity management solution however.  I will contrast them with strategic capacity management solutions on Monday to illustrate why just taking one approach, strategic or operational, is not enough.

In the meantime don't forget to register for our 'Data Correlation in Capacity Management' webinar

Andrew Smith
Chief Executive Officer

Wednesday, 8 July 2015

Capacity Management: Operational or Strategic? (1 of 4)

Many software suppliers seem to be talking about how they do capacity management these days.  
In the server area of infrastructure management alone you have:

·        Established businesses that have been offering data capture, capacity database and a    range of capacity reporting for some time.   Typically these businesses have started in one technology area such as mainframe or Unix and spread their coverage across platforms as IT has evolved.

·        Point solutions that analyse current performance and make tuning recommendations

·        Framework providers touting capacity management as one of the modules within their toolset.  Often this is the result of acquisition of a third party capacity management product that then gets increasingly integrated with their framework

·        SaaS/Cloud solutions that take in data from one or many environments

In this blog, I’d like to consider the first two of these options.

Common with environments that support rapid provisioning, is confusion between the first and second areas.  Both groups talk of capacity management, but the focus of what capacity is being managed is different.  This is not the metrics or applications that are being watched, more the time frame and nature of capacity events that are being reviewed.

In days gone past, one might have talked of this as the difference between capacity management and performance management.  As everyone seems to be using the terminology much more interchangeably these days, perhaps it is better expressed as strategic capacity management and operational capacity management.

Operational capacity management is based on measured performance metrics and recommends operational changes, e.g. moving VMs.

Strategic capacity management is based on taking action well in advance to remove the risks and problems before they occur.

I’ll be taking a look at both of these types of capacity management on Friday, in the meantime why not sign up to our Community and access our range of capacity management white papers and webinars

Andrew Smith

Chief Executive Officer

Thursday, 2 July 2015

Data Correlation for Capacity Management

Correlation is used across many disciplines to identify predictive relationships that can be used in decision support. 

Correlating capacity and performance data is an important tool that analysts should be well versed in. Many software applications are available to assist the analyst in finding correlations and identifying the significance of those dependencies. 

A classic example is correlating workload volumes to resource consumption when calibrating models. Many types of data can be correlated to gain insight into what drives resource utilization and performance throughout the entire computing environment.

I'll be running a webinar which presents a high-level discussion of using correlation in practice and doesn't attempt a rigorous mathematical explanation of the underlying statistics. A rigorous mathematical review can be found on line at many websites with an academic focus for those readers who are interested. 

I'll be reviewing basic concepts of correlation and looking at significance coefficients,

along with limitations of correlation, the types of data to correlate and I'll be showing you some examples. 

I intend to give readers a better working knowledge of how correlation can be used in practice to make informed decisions regarding their capacity and performance management.
Join my webinar on July 22, register for your place now.

Dale Feiste
Principal Consultant